We estimated the mortality from various diseases caused by cigarette smoking using two methods and compared the results. In one method, the "Prevent" model is used to simulate the effect on mortality of the prevalence of cigarette smoking derived retrospectively. The other method, suggested by R. Peto et al (Lancet 1992;339:1268-1278), requires data on mortality from lung cancer among people who have never smoked and among smokers, but it does not require data on the prevalence of smoking. In the Prevent model, 33% of deaths among men and 23% of those among women in 1993 from lung cancer, chronic bronchitis, emphysema, ischemic heart disease, and stroke were caused by cigarette smoking. In the method proposed by Peto et al, 35% of deaths among men and 25% of deaths among women from these causes were estimated to be attributable to cigarette smoking. The differences between the two methods are small and appear to be explicable. The Prevent model can be used for more general scenarios of effective health promotion, but it requires more data than the Peto et al method, which can be used only to estimate mortality related to smoking.